A Novel Reduced Complexity Trellis-based MAP Decoding Algorithm for Block Turbo Codes

In this paper ure present a new reduced complexity trellis-based maximum a posteriori probability (MAP) decoding algorithm for Block Turbo Codes (BTCs). To reduce the complexity, a simplified trellis is derived at the end of each iteration in which many branches and states are removed. Trellis reduction is accomplished based on the statistical properties of the log-likelihood ratio (LLR) information. The performance of Bose-Chaudhuri-Hocquengham (BCH)-BTC over additive white Gaussian noise (AWGN) channel using the proposed decoding algorithm is evaluated. Simulation results reveal that the innovative strategl of trellis reduction has negligible bit-error rate (BER) degradation in comparison with the full-search log-MAP algorithm.